Oncogene (2003) 22, 73597368

& 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00
www.nature.com/onc

Rituximab (monoclonal anti-CD20 antibody): mechanisms of action and

resistance
Mitchell R Smith*,1
1

Lymphoma Service, Fox Chase Cancer Center, 7701 Burholme Avenue, Philadelphia, PA 19111, USA

Rituximab, a chimeric monoclonal antibody targeted

against the pan-B-cell marker CD20, was the rst
monoclonal antibody to be approved for therapeutic use.
Treatment with rituximab at standard weekly dosing is
effective in more than 50% of patients with relapsed or
refractory CD20-positive follicular non-Hodgkins lymphoma, but is not curative. It is less effective in other
subtypes of CD20-positive lymphoma and for retreatment,
even with CD20 still expressed. Thus, binding of
rituximab to CD20 is not sufcient to kill many
lymphoma cells, indicating that there are mechanisms of
resistance. Mechanisms of cell destruction that have been
demonstrated to be activated by rituximab binding to
CD20 include direct signaling of apoptosis, complement
activation and cell-mediated cytotoxicity. The relative
importance of each of these mechanisms in determining
clinical response to rituximab treatment remains a matter
of conjecture. Thus, the role of various resistance pathways, some documented in experimental systems and
others still hypothetical, remains uncertain. Resistance
could potentially be mediated by alterations in CD20
expression or signaling, elevated apoptotic threshold,
modulation of complement activity or diminished cellular
cytotoxicity. As the rst of an expanding class of
anticancer agents, lessons learned regarding the mechanism of rituximab action and resistance will be of
increasing importance.
Oncogene (2003) 22, 73597368. doi:10.1038/sj.onc.1206939
Keywords: antibody dependent cellular cytotoxicity;
non-Hodgkins lymphoma; B-cell receptor signaling;
apoptosis

Introduction
Rituximab, a chimeric monoclonal antibody that binds
to CD20, was the rst monoclonal antibody to be
approved for clinical use in the therapy of cancer. It is
approved for use against indolent B-cell non-Hodgkins
lymphoma (NHL), although its use has expanded
signicantly beyond that indication to virtually any
CD20-positive NHL, and more recently into other areas
such as autoimmune disorders. This review will focus on
*Correspondence: MR Smith; E-mail: M_smith@fccc.edu

efcacy of, and resistance to, rituximab in treatment of

B-cell NHL.
The search for tumor-specic antigens that would
serve as targets for antibody-mediated attack on cancer
cells has been largely disappointing to date, although
genomic and proteomic approaches hold out the
promise of the future identication of such tumorspecic proteins. NHL are malignancies of lymphocytes.
About 8090% express B-cell markers and have
immunoglobulin gene rearrangements indicating their
clonal derivation from a B-cell progenitor, while most of
the remainder express T-cell antigens and harbor clonal
T-cell receptor gene rearrangements. Given the preponderance of B-cell malignancies, attention was
focused on these in monoclonal antibody development.
Since clonal immunoglobulin gene rearrangements
translate into a clonal surface immunoglobulin specic
for the NHL cell, these immunoglobulins are truly
tumor-specic antigens. Thus, anti-idiotype antibodies
had potential as tumor-specic therapeutic agents.
While such anti-idiotype antibodies have clinical activity, the logistics of developing an individual therapeutic
antibody for each patient proved insurmountable
(Grillo-Lopez, 2000). B-cell-specic markers could be
similarly targeted and would not require individualized
development. Rituximab targets CD20, a transmembrane protein present on virtually all B cells from the
stage at which they become committed to B-cell
development until it is downregulated when they
differentiate into antibody-secreting plasma cells (Reff
et al., 1994). Thus, CD20 is considered a pan-B-cell
antigenic marker. The role of CD20 in B-cell development remains uncertain, as CD20-decient mice have
normal B cells (OKeefe et al., 1998). Similarly, CD20 is
expressed on virtually all B-cell NHL, but not on most
multiple myeloma cells that correspond to more mature
B cells with plasma cell differentiation (Anderson et al.,
1984). Targeting such a marker, however, would be
expected to be associated with the confounding problem
that it would not only kill lymphoma cells, but also
normal B-cell counterparts. In fact, rituximab does clear
both normal and malignant CD20-positive cells, but the
absence of normal B cells for approximately 6 months
has not been associated with decrease in IgG levels or
signicant increase in infectious risk (McLaughlin et al.,
1998). Other major considerations in selecting CD20 as
an appropriate target included that the antigen is
expressed at reasonably high levels, is not downregu-

Rituximab, mechanisms of action and resistance

MR Smith

7360

lated after antibody binding and is not shed or secreted

into the circulation, which avoids the antibody binding
solely to these circulating molecules and not reaching
the NHL cell (Grillo-Lopez, 2000).
NHL is an increasing problem, now being the fth
most common malignancy in the United States (Greenlee et al., 2000). Since the incidence rises with age, as the
population ages NHL has become increasingly common,
but there has also been over the past 40 years an
increased incidence at any given age. The reason for the
increase is not clear, but attention has focused on
environmental factors. The histopathologic classication of NHL has historically been quite complex.
Improved understanding of the biology of normal Bcell development, and that B-cell NHL correspond to
specic stages of normal B-cell development, has led to a
more rational classication of B-cell malignancy (Harris
et al., 1994). About 1/3 of NHL have a follicular growth
pattern, reminiscent of normal lymph node architecture,
and these generally have an indolent clinical course, or
low-grade histology. Other less common indolent B-cell
NHL are small lymphocytic lymphoma, with cells
indistinguishable from chronic lymphocytic leukemia,
and marginal zone- or mucosa-associated NHL.
Patients with these indolent NHL have a median
survival of approximately 10 years (Armitage and
Weisenburger, 1998), but the disease is considered
incurable with current therapeutic options. The usual
clinical course is a series of remissions and relapses, with
different treatments used for each relapse. Thus, new
treatments are needed. About 1/3 of NHL are diffuselarge-cell or immunoblastic NHL with a more aggressive
clinical course. These are potentially curable with
combination chemotherapy, but if not cured usually
are fatal within 23 years (Fisher et al., 1993). Again,
improved therapy is needed.
Patients with indolent NHL were considered a
suitable population in which to test antibody-based
therapy initially, because there is often no urgent need
for therapy. Since antibody-based therapy might be slow
to act, an indolent disease would also permit a safe
period of observation. Rituximab is an active therapy as
a single agent against indolent NHL. The success of
rituximab likely relates to a number of factors, one of
which is that it is a chimeric rather than a murine
antibody (Grillo-Lopez, 2000). Rituximab retains the
murine CD20-binding Fab regions, but uses a human Fc
portion (Figure 1). This has at least two major effects.
First, with less anti-mouse antibody (HAMA) development, the t1/2 of the antibody is prolonged by the
presence of the human Fc portion. Second, the Fc
portion contains the effector aspects of the molecule,
including complement activation and attraction of
cytotoxic cells, so that the human Fc is more effective
in activating these effectors in patients.
Rituximab efcacy and scope of clinical resistance
Rituximab is approved for use in refractory or relapsed
indolent NHL, based on a pivotal trial demonstrating a
Oncogene

Figure 1 Rituximab chimeric structure. Binding regions from

original murine anti-human CD20, consisting of variable regions of
immunoglobulin heavy and light chains, are fused to human IgG1
heavy-chain and human kappa light-chain constant regions. Fc
portion from human IgG1 was selected for its ability to x
complement and activate antibody-dependent cellular cytotoxicity

response rate of over 50% in such patients (McLaughlin

et al., 1998). The responses in this study lasted just over
1 year. The standard dose and schedule for rituximab,
used in this trial, is 375 mg/m2 given once weekly for 4
weeks. In phase I trials, no true maximum tolerated dose
was reached due to lack of toxicity, so the dose and
schedule were selected somewhat empirically and limited
by practical considerations about rituximab availability.
Thus, higher doses, more frequent administration and/
or additional doses are now possible. Owing to its
efcacy and tolerability, trials of rituximab as induction
therapy are underway and preliminary results suggest
higher responses than in relapsed patients (Colombat
et al., 2001; Hainsworth et al., 2002), although
ultimately improved survival is the critical end point.
While rituximab is an effective addition to therapy for
CD20-expressing NHL, it is active in patients with
relapsed or refractory disease in only 5060% of
follicular lymphoma and 1015% of small lymphocytic
lymphoma (McLaughlin et al., 1998), despite CD20
expression in all cases treated. Furthermore, patients
with indolent NHL treated with rituximab rarely
develop CD20-negative disease. Yet, of those who
respond to rituximab and then relapse more than 6
months later, only 40% will respond again when
retreated with rituximab (Davis et al., 2000). Thus,
there must be mechanisms of resistance independent of
CD20 expression.
Evidence for multiple mechanisms of rituximab action
has been reported, and it remains unclear which is/are
most important in patients. Therefore, it is difcult to
know the relative importance of potential mechanisms
of resistance. What is known about the steps that lead to
rituximab activity, as well as potential levels of
resistance, is reviewed here, with the view that this
schema will serve as a framework to consider potential
resistance mechanisms (Table 1) and to highlight specic
areas that require attention in clinical trials. The steps
are divided into those that lead up to CD20 binding to
the surface of the NHL cell, that is, pre-CD20 binding,
and the results of such binding that lead, hopefully, to
NHL cell elimination, that is, post-CD20 binding. Many

Rituximab, mechanisms of action and resistance

MR Smith

7361
Table 1 Summary of potential mechanisms of rituximab action, corresponding means of resistance and conceptual approaches that might address
these resistance mechanisms. These issues are discussed further in the text
Mechanism of action
CD20 binding

Examples of resistance mechanisms

k/absent CD20 expression
m accessible CD20
Soluble CD20
m tumor burden
m Ab metabolism
Poor tumor penetration

Post-CD20 binding
B-cell signaling
Direct apoptosis
Complement activation
ADCC

Membrane rafts/clustering
Altered PLC
Altered tyrosine kinase
Elevated bcl-2
Complement inhibitors
FcgRIIIA polymorphism
k Effector cells (e.g. after chemotherapy)
Inhibit perforin/granzyme

of the pre-CD20-binding considerations are similar to

traditional drugs, while others depend on the specicity
of binding and the distribution of target sites. PostCD20 binding effects may include receptor signaling,
calcium ux, direct apoptosis, complement activation or
attraction of antibody-dependent effector cells. Potential exists for resistance at any one of these steps.

Events up to CD20 binding

Many events occur prior to rituximab binding to tumor
cell surface CD20. These events determine whether
rituximab reaches a sufcient percentage of the lymphoma cells in sufcient quantities to lead to cytotoxicity. Sufcient, however, remains at best empirically
dened.
Serum levels of rituximab
As a targeted monoclonal antibody, rituximab behaves
differently from standard chemotherapeutic agents, in
that its pharmacokinetics are inuenced not only by the
characteristics of an antibody molecule, but also by the
presence of the target, and whether the target is
circulating or cellular. The half-life of a monoclonal
antibody in human circulation is determined, in part at
least, by the Fc portion of the immunoglobulin
molecule. The actual level, however, depends upon the
total amount of accessible CD20, a value that reects
circulating CD20 and both the number of CD20positive cells and the density of CD20 per cell. Levels
of rituximab tend to increase with repeated doses
(Berinstein et al., 1998) as the number of CD20
molecules decreases. Serum rituximab levels have been
reported to be higher in responders than in nonresponders (Berinstein et al., 1998). However, whether the
higher levels lead to a better response rate or whether
responders have fewer CD20 molecules due to lower

Potential means to overcome resistance

Higher/more frequent rituximab doses
Cytokines to m CD20 expression
Cytoreductive therapy+rituximab
Administer with IVIg
Alter afnity, charge or size of Ab

Small molecule
Kinase activators/inhibitors
Antisense bcl-2
Fludarabine to k CD55/CD59
Antibodies to CD55/CD59
Ab engineered to bind to FcgRIIIA F158
Restore effector cells (e.g. cell transfer or cytokine expansion)
Small molecules to block inhibitors

tumor burden and therefore higher circulating antibody

levels have not been determined. Nonetheless, rapid
metabolism of rituximab, due either to high numbers of
accessible CD20 molecules and/or to alterations in host
antibody metabolism, could be a basis for resistance.
Attempts to circumvent this possibility with repeated
dosing have been pursued to raise antibody levels and
therefore to try to improve response rates. Use of 8
weekly doses rather than the standard 4 weeks in a
nonrandomized trial suggested the possibility of a higher
response rate in relapsed/refractory low-grade lymphoma (Piro et al., 1999). In CLL, a disease in which the
cells have low CD20 levels and a low response rate to
standard dose and schedule rituximab (375 mg/m2
weekly
4), higher and repeated doses did lead to
higher response rates, although these tend to be short
lived (Byrd et al., 2001; Keating et al., 2002b). Another
approach would be to individualize dosing to maintain
rituximab levels above a yet to be determined threshold
for prolonged periods. In fact, a randomized trial has
shown a benet for additional rituximab doses every 2
months for four doses after the initial 4-week cycle
(Ghielmini et al., 2002). Alternatively, since normal B
cells begin to recover about 6 months after rituximab
administration, a schedule of 4 weeks of rituximab every
6 months has been used (Hainsworth et al., 2002).
Development of anti-chimeric antibodies (HACA) that
might rapidly reduce rituximab levels has been an
uncommon nding in clinical trials, likely reecting the
human Fc portion as well as the generally immunosuppressed status of the patients.
Rituximab distribution
Antibody distribution in various compartments within
the body, as well as within a malignant lymph node or
extranodal tumor mass, may affect drug sensitivity or
resistance. This concept is most clearly illustrated by
data for the monoclonal anti-CD52 antibody CampathOncogene

Rituximab, mechanisms of action and resistance

MR Smith

7362

1H, found to be much more efcient at clearing NHL

cells from blood and marrow than from lymph nodes,
leading to studies of this agent in CLL (Osterborg et al.,
1997). In the CLL studies, again responses are higher in
blood and marrow than in nodes, although with not as
marked a difference as in NHL (Osterborg et al., 1997;
Keating et al., 2002a). Similarly, encouraging reports of
the ability of rituximab therapy to clear blood and
marrow of lymphoma cells, even as assayed by sensitive
PCR methods for IgH gene rearrangements or t(14;18),
were followed by the realization that even patients with
molecular remissions continue to relapse, indicating
persistence of lymphoma cells (Foran et al., 2000;
Pichert et al., 2001). While more complete responses
may translate into longer disease-free intervals, patients
are not cured (Czuczman et al., 2001). Thus, antibodies
may be more successful in removing cells from blood
and marrow than from the rest of the body. Such
ndings could reect differential access to antibody or
to effector mechanisms (see below). Little data exist on
the access of rituximab or other therapeutic antibodies
to traditional chemotherapy sanctuary sites, such as
brain or testis, which are protected from toxins by
physiologic barriers, although one case report detected
only very low levels of rituximab in the CSF of a patient
with CNS involvement of systemic NHL (Harjunpaa
et al., 2001).
Other factors need to be considered regarding
distribution of antibody within lymph nodes and
lymphoma. IgG molecules move readily out of the
vessels into the tissue. The distance from a vessel
reached by effective concentrations of antibody depends
on the diffusion characteristics of the antibody molecule, in turn largely determined by size and charge, as
well as the afnity of the antibody for its target and the
target concentration in tissue (Figure 2). Rituximab has
an apparent afnity constant of 5.2 nm for CD20 on
human SB cells (Reff et al., 1994). Thus, while it might
seem reasonable to strive for the highest afnity when
engineering antibodies, too high an afnity and high
numbers of target molecules can lead to antibody being
bound only to targets close to the vessel, but not
penetrating more deeply into the tissue. Too low an
afnity and binding may be inadequate to have an
effect. Smaller constructs, such as single-chain Fv
fragments, have been designed to improve tissue
penetration, but these need to be optimized based on
afnity as well.
Surface CD20 expression
CD20 expression is quite heterogeneous in different
lymphoma types, as well as among cells of an individual
tumor sample. This can be visualized when examining
the distribution in a ow cytometric histogram of a
suspension of lymphoma cells stained with anti-CD20
(Figure 3). A positive sample is typically determined as
having 430% of cells above the cutoff, which varies
with the staining of the isotype control antibody. Thus,
even in a CD20-positive lymphoma, many cells may
have low or background staining, and the intensity of
Oncogene

Figure 2 Potential effects of antibody characteristics and target

antigen distribution on delivery of antibody as it diffuses from the
vessel into the tissue. Tissue penetration of antibody depends on
target density, antibody afnity and diffusion characteristics of the
antibody. The latter depends on physical factors such as size,
charge and aggregation. The width of the triangles indicates
concentration of the antibody at variable depth of penetration
from the blood vessel. The optimal balance for sufcient antibody
delivery to an adequate depth of penetration is not certain

Figure 3 Flow cytometric analysis of CD20. Schematic histogram

demonstrates that surface marker expression has a normal
distribution. Some cells will have low expression with a signal that
overlaps that of negative cells. The demarcation point between
positive vs negative depends on the characteristics of the negative
control and positive cell populations and is somewhat arbitrary, as
is the number of cells that must be above this level to call a
specimen positive. CLL/small lymphocytic lymphoma cells typically have dim CD20 expression, so that, while they may be clearly
positive by visual inspection of the proles, they may be considered
negative. Note that the abcissa is a log scale, so that there is a wide
range of levels of expression for individual cells within a given
sample

the positive cells may vary over a log range. Typically,

CLL and small lymphocytic lymphoma have dim CD20
staining, and a corresponding lower rituximab response
rate than follicular lymphoma, at least for previously
treated disease (McLaughlin et al., 1998). Mantle-cell
and diffuse-large-cell lymphoma, however, have similar
or even higher CD20 expression than follicular lymphoma,

Rituximab, mechanisms of action and resistance

MR Smith

7363

but lower response rates, so that staining intensity alone

does not predict for response. Furthermore, if CD20
expression alone determined rituximab responsiveness,
then one would expect development of CD20-negative
relapses after rituximab treatment, and this circumstance has rarely been reported, at least for indolent
lymphoma (Davis et al., 2000). Recent reports are
perhaps suggestive that CD20-negative relapses may
develop in aggressive lymphomas (Davis et al., 1999;
Foran et al., 2001; Kennedy et al., 2002), but care must
be taken to differentiate true CD20 negativity from
false-negative results attributable to blocking of labeled
CD20 antibody by bound rituximab. There have been
attempts to upregulate CD20 to try to enhance
rituximab efcacy, such as with G- or GM-CSF, but
mechanisms other than CD20 upregulation may account
for any enhanced effect (Ravetch and Lanier, 2000; van
der Kolk et al., 2002).

Events after CD20 binding

The events that lead to cell killing following antibody
binding to CD20 may be multifactorial. These events
undoubtedly inuence the cytotoxicity of rituximab and
resistance mechanisms in that there are tumor cells that
bind rituximab but are not killed. Thus, there must be
postbinding mechanisms of resistance.
CD20 binding as a B-cell signal
While the exact role of the transmembrane CD20
molecule is not known, it is involved in B-cell
differentiation and activation (Golay et al., 1985;
Tedder et al., 1985), although CD20 knockout mice
have normal B-cell number and development (OKeefe
et al., 1998). CD20 can act as a calcium channel (Bubien
et al., 1993), either directly or by binding to or activating
a calcium channel. Binding by rituximab, especially if
crosslinked, initiates a cascade of intracellular signals
(Figure 4). These signals may play a role, at least in part,
in rituximab-mediated cell killing. CD20 is associated
with a number of protein tyrosine kinases, including lyn,
fyn, lck and p75/85 kinase (Deans et al., 1995). CD20
engagement leads to activation of phospholipase Cg
(Deans et al., 1993). Inhibitor studies indicate that this
occurs via src-family kinases (Shan et al., 2000). The
pleiotropic effects of PLC-g including MAP kinase
activation and specically JNK, ERK and p38MAPK
have been implicated in rituximab signaling (Pedersen
et al., 2002). PLC-g also cleaves PIP3, generating inositol
triphosphate and a resultant calcium ux, the latter
being implicated in CD20-mediated apoptosis by the
observation that intracellular and extracellular calcium
chelation each inhibited apoptosis induced by antiCD20 antibody in RAMOS cells (Hofmeister et al.,
2000). Calcium ux could also relate to the direct effects
of CD20 as a calcium channel. PLC-g cleavage of PIP3
also generates diacylglycerol with resultant protein
kinase C (PKC) activation, although this pathway has
not yet been explored as a potential mechanism for anti-

Figure 4 Outline of intracellular signaling pathways reported to

be activated by crosslinking of CD20. Listed enzymes are those for
which data have been published, although clearly an array of other
intersecting pathways could be involved and need to be explored.
Note the similarities of CD20 signaling to BCR signaling pathways,
each of which involves lipid rafts

CD20 activity. Many of these signaling pathways have

been found to be quite similar to signaling after
physiologic engagement of the B-cell receptor (BCR)
(Mathas et al., 2000). The result of stimulation of these
complex signaling pathways, that is, cell cycle arrest or
apoptosis, appears to be cell line specic. One can
certainly envision that multiple factors including, but
not limited to, the growth state of the cell, presence of
costimulatory surface molecules and their ligands and
simultaneous signals would determine whether the
response of a cell to CD20 binding would be none,
growth arrest, growth stimulation or apoptosis. Resistance to rituximab killing, therefore, might be due to as
yet undened alterations in intracellular signals induced
by CD20 binding.
The extent of intracellular signal transduction, at least
in some systems, can also be affected by the clustering of
CD20 (Hofmeister et al., 2000). This may be dependent
on the density of CD20 and the amount of antibody
present. Another means of clustering is the nding that
after CD20 is bound, it rapidly redistributes into a cell
membrane fraction called a lipid raft (Deans et al., 1998,
2002), which may aid in signal activation through the
src-family tyrosine kinases. These membrane changes
may be long lived, and are one of several potential
explanations for late responses to antibody therapy. Of
interest is that CD20 associates with the BCR in these
lipid rafts (Petrie and Deans, 2002), suggesting that the
observed similarities in BCR and CD20 signaling may
have a structural basis.
Clustering of CD20 has been more directly related to
crosslinking, that is, crosslinking of the antibody
molecules by several different mechanisms amplies
signaling. Thus, whereas slight tyrosine kinase activaOncogene

Rituximab, mechanisms of action and resistance

MR Smith

7364

tion and calcium ux may be observed after soluble antiCD20 binds to cells, quantitatively more pronounced
signals develop when homodimerized antibody is used
(Ghetie et al., 2001) or when antibody is crosslinked by
secondary antibody (Cragg et al., 2003) or by cells
expressing Fcg receptors (Shan et al., 1998). The
potential role of activating and inhibitory receptors of
the Fcg family is discussed below; however, the
inhibitory FcgRIIB (CD32) is the Fcg receptor on B
cells. Binding of the Fc portion of IgG to CD16 blocks
simultaneous BCR-mediated signaling (Ravetch and
Lanier, 2000). Given the similarities between BCR and
CD20 signaling noted above, complex signaling effects
downstream of rituximab binding to CD20 may occur.
Also of interest is that CD22 is also a B-cell specic
inhibitory signaling receptor, suggesting a possible
mechanism of interaction between rituximab and antiCD22 monoclonal antibody, a combination that has
shown promising activity in clinical trials (Leonard and
Link, 2002). Adjacent lymphoma cells themselves may
act as crosslinking agents if they express CD32 or other
Fc receptors. The degree of in vivo crosslinking and its
importance to rituximab activity remains to be elucidated.
CD20 binding induces apoptosis
While cell- and complement-mediated cell death is
discussed later, there is evidence, at least in some
experimental systems, that cell death is mediated by
antibody through apoptotic pathways (Figure 5). Rituximab-mediated apoptosis is thought to be a consequence
of caspase-3 activation, and recent data from patients
with CLL support this concept (Byrd et al., 2002),
whereas the FAS ligand/FAS death pathway does not
appear to be necessary. Thus, the mitochondrialdependent pathway would be likely to be important.
However, the role of bcl-2-dependent pathways remains
unclear. In the EBV-positive 2F7 cell line derived from
AIDS-associated Burkitts lymphoma, rituximab binding led to decreased bcl-2 expression and sensitization to
chemotherapy acting via mitochondrial pathways (Alas
et al., 2002). This was shown to be due to downregulation of IL-10, resulting in decreased activation of
STAT3 through an IL-10/IL-10R autocrine loop. In
contrast, EBV-infected bcl-2-expressing RAMOS-AW
and the EBV-negative bcl-2-nonexpressing parent
RAMOS are equally sensitive to anti-CD20 apoptosis
(Shan et al., 2000), and no signicant alterations in bcl-2
levels were seen after anti-CD20 treatment. One
consideration for the relative insensitivity of CLL to
rituximab, in addition to dim CD20 expression, is that
bcl-2 is overexpressed in these patients raising the
apoptosis threshold. However, follicular NHL in which
bcl-2 is dysregulated by t(14;18), is very sensitive to
rituximab, so that complex factors must regulate
resistance vs sensitivity. Recent data suggest that downregulation of bcl-2 by antisense oligonucleotides does
enhance rituximab efcacy (Smith et al., submitted), but
whether this involves the two agents affecting the same
or parallel pathways needs to be further elucidated.
Oncogene

Figure 5 Apoptotic control pathways involved in rituximabinduced apoptosis. The IL-10 autocrine loop has been identied in
the 2F7 cell line, but not in others, as downregulating bcl-2. XIAP
and mcl-1 have been reported to play a role in the control of
rituximab-induced apoptosis in patients with CLL. The FAS
ligand/TNF/TRAIL pathway does not appear to be involved in
CD20-induced cell death. Any alterations in signals that affect proand antiapoptotic balance could potentially alter rituximab
sensitivity

While little data yet exist to prove or disprove the

hypothesis, alterations in the set-point for apoptosis,
perhaps by changes in the levels of bcl-2 or its related
family of proteins, apoptosome components such as
APAF-1, or other regulators of apoptosis sensitivity
could lead to rituximab resistance allowing rituximabcoated cells to escape cell death.
In general, there has been wide variation in the
amount of direct apoptosis observed, and this may
reect differences in cell lines, growth conditions, source
and Ig subclass of anti-CD20 antibody, and different
time points of analysis. The question of the relevance of
studies of rituximab effects on Burkitts lymphoma cell
lines must be kept in mind, as this is not the clinical
situation in which the drug has been tested. Little data
on indolent lymphoma cell lines, corresponding to the
predominant clinical use, have been reported, although
transformed lymphoma cell lines have been tested and
some are quite sensitive to direct apoptosis. Recent data
from patients with CLL treated with rituximab do,
however, support a role for intravascular apoptosis
through the mitochondrial pathway, that is, activation
of caspases 3 and 9, but not caspase 8, and downregulation of XIAP and mcl-1 (Byrd et al., 2002).
Role of complement activation in CD20 efficacy
Complement activation by the Fc portion of the
antibody leading to cell lysis, or complement-dependent
cytotoxicity (CDC), is another postulated mechanism of

Rituximab, mechanisms of action and resistance

MR Smith

7365

action of monoclonal anti-CD20 (Figure 6). As for other

potential mechanisms of antibody activity, in vitro
evidence conrms that this phenomenon exists, but
there is signicant variability in sensitivity among B-cell
lymphoma cell lines. In vitro testing of cells from
patients has correlated CD20 levels with sensitivity to
rituximab-mediated CDC in some reports (Golay et al.,
2001; Bellosillo et al., 2001), but not others (Manches
et al., 2003). In vitro CDC, however, did not correlate
with clinical response (Weng and Levy, 2001).
Complement lysis is controlled not only by the degree
of activation, but also regulated by a series of
complement inhibitory proteins, especially CD35 (complement receptor type 1, CR1), CD46 (membrane
cofactor protein), CD55 (decay accelerating factor)
and CD59 (membrane inhibitor of reactive lysis,
protectin) (Charles and Foerster, 1999). CD59 was
identied on rituximab-bound persistent cells in patients
(Treon et al., 2001), suggesting a possible role in
acquired resistance. While evaluation of a panel of cell
lines, many of which were myeloma cells, suggested that
CD59 expression correlated with CDC (Treon et al.,
2001), expression of these proteins has not been found in
patient samples to predict in vitro CDC (Golay et al.,
2001) or clinical response (Weng and Levy, 2001).
Nevertheless, blockade of CD55 and CD59 has been
reported to enhance rituximab-mediated CDC (Golay
et al., 2000, 2001). Also, the nucleoside analog
udarabine, clinically active against CLL and indolent

Figure 6 Mechanism of antibody-dependent complement lysis

and potential resistance mechanisms. The complement cascade
involves a series of zymogens that are sequentially activated, as well
as a series of inhibitors (indicated in bold). Two closely
approximated immunoglobulin Fc portions initiate the classical
pathway by activating C1, which in turn activates C4 and C2. This
reaction is closely controlled by circulating C1 inhibitor. The C1,
C4 and C2 complex, known as C3 convertase, activates C3 leaving
a fragment, C3b, on the cell surface. C3b associating with the
plasma membrane becomes C5 convertase. Membrane-associated
decay accelerating factor (DAF, CD55) and complement receptor 1
(CR1, CD35) inhibit both C3 and C5 convertase. C5b, C6, C7, C8
and C9 then associate to form the MAC. This activity is inhibited
by the membrane inhibitor of reactive lysis (CD59). Note that
CD55 and CD59 are linked to the membrane by glycosyl
phosphotidylinositol or GPI. Red blood cells decient in GPIlinked molecules are exquisitely sensitive to complement-mediated
lysis in the disease paroxysmal nocturnal hemoglobinuria (PNH),
demonstrating that complement sensitivity of cells can be
controlled by modulating the cell surface. Furthermore, cells
opsonized by IgG and C3b molecules remaining on the surface are
targets for phagocytosis via Fc receptors or the C3b receptors CR1
(CD35). It is interesting to note that red cells targeted by high-titer
IgG are largely cleared extravascularly by the reticuloendothelial
system rather than by intravascular direct complement lysis.
Clearly, there are many levels of potential control of cell
destruction after antibody binding

NHL, has been reported to downregulate CD55 and

CD59 (Di Gaetano et al., 2001) and this is a potential
means by which udarabine rituximab may have
benecial clinical efcacy. A recent report suggests,
based on data with nonsaturating levels of anti-CD20,
that an overall assessment of the levels of both CD20
and the complement inhibitory proteins can predict in
vitro sensitivity to rituximab-mediated CDC, and that
this may correlate with clinical response (Manches et al.,
2003).
It appears that the ability of CD20 to move into lipid
rafts is a requirement for CDC to occur (Cragg et al.,
2003). Whether this is necessary to activate the early
stages of the complement pathway or somehow increases the sensitivity to the membrane attack complex
(MAC) needs additional study. In addition, a role has
been reported for the protein kinases PKC, PKA and
MEK in complement resistance of some tumor cells
(Donin et al., 2003). Thus, inhibitors of these kinases
sensitized the carcinoma cells to complement lysis.
Perhaps, we can also obtain insight into complementassociated resistance mechanisms from parasites that
have evolved the capacity to evade this host system
(Sacks and Sher, 2002).
It is clear that complement activation occurs in
patients treated with rituximab and, in fact, seems to
be related to many of the infusional toxicity reactions
associated with this agent. Protein fragments indicative
of complement activation have been measured in
patients. The relative importance of this as a clearance
mechanism, however, remains controversial. Further,
complement activation has other effects besides lysis by
the nal MAC, such as depositing C3, C3b and
additional C3b breakdown products on the cell surface.
Membrane-attached C3 and C3 fragments could be
targets of C3 receptor on effector cells, enhancing cell
killing by antibody-dependent cellular cytotoxicity
ADCC (see below) (Kennedy et al., 2003). Thus, the
precise role of complement activation and potential as a
mechanism of rituximab resistance requires additional
investigation.
Antibody-dependent cellular cytotoxicity
ADCC in the presence of rituximab represents killing of
lymphoma cells by effector cells that are activated by
binding to the Fc portion of the chimeric anti-CD20
molecule coating the lymphoma cell (Figure 7). Members of the Fcg receptor family are expressed on
monocytes, macrophages and granulocytes and include
the activating high-afnity FcgRI (CD64) and lowafnity FcgRIIIA (CD16), as well as the inhibitory lowafnity FcgRIIB (CD32). The activating FcgRIIIA
(CD16) is also present on NK cells, while the inhibitory
FcgRIIB (CD32) is a key regulator on B lymphocytes
(Clynes et al., 2000; Ravetch and Lanier, 2000). Two
recent avenues of investigation have pointed to ADCC
as an important in vivo mechanism of rituximab action,
as well as other antibodies. First, knockout mice
decient in the inhibitory FcgRIIB are hypersensitive
to antibody-mediated tumor suppression (Clynes et al.,
Oncogene

Rituximab, mechanisms of action and resistance

MR Smith

7366

Figure 7 Cellular cytotoxicity mechanisms and controls after

rituximab binds CD20 on B-cell surface. NK cell cytotoxicity and
cytokine secretion in response to a rituximab-coated B cell is
determined by the balance of activating and inhibiting signals
through HLA-restricted receptors. Inhibitory signals are via ITIM
recruiting SHP-1. Activation is via an adapter, DAP12, that
contains ITAM. For myeloid cells, the balance of signals through
activating Fcg receptors CD16 and CD64 and inhibitory Fcg
receptor CD32 determines whether various cytotoxic mechanisms
are induced when the cell encounters a rituximab-coated B cell

2000). In mice with the common g chain knocked out,

which are decient in both activating receptors (FcgRI
and FcgRIII), antibodies were no longer effective in
inhibiting tumor growth. Further, antibodies engineered
to have reduced ability to interact with the FcgR were
unable to suppress tumor growth. These data indicate
that, at least in nude mouse xenograft models, antibody
control of tumor growth requires FcgR interactions, and
that the balance of activating and inhibitory receptor
signaling is important. Second, evidence that FcgR plays
a direct role in efcacy in patients treated with rituximab
comes from investigations of FcgRIIIA genetic polymorphisms. A dimorphism exists in which residue 158 of
FcgRIIIA can be either valine, with stronger binding to
IgG1, or phenylalanine with weaker binding (Koene
et al., 1997; Wu et al., 1997). In a small group of patients
with previously untreated follicular lymphoma who
received rituximab, response rates and molecular
complete remission rates were higher in 10 of the 49
patients homozygous for the FcgRIIIA with V158, than
those who had either one or two copies of FcgRIIIA
with F158 (Cartron et al., 2002). These data not only
indicate a key role for FcgRIIIA in rituximab activity,
but also suggest that response may be predicted and
more effective antibody constructs engineered to overcome this basis of resistance.
Receptors other than FcgR may also be important.
NK cells are normally inhibited from attacking MHC-1
identical cells, that is, self, by coligation of MHC-1
with inhibitory killer Ig-like receptors or KIR. Additional activating and inhibitory coligating signals exist
and the balance determines NK activity (Campbell and
Colonna, 2001). These signals are transmitted from the
membrane into the cell by transmembrane proteins
containing immunoreceptor tyrosine-based activation or
inhibitory motifs (ITAM and ITIM). The inhibitory
Oncogene

FcgRIIB and KIR signal via ITIM domains that are

substrates for src-family tyrosine kinases and that in
turn activate PIP3 phosphatase (SHIP) for FcgRIIB or
tyrosine phosphatase (SHP-1, -2) for KIR (Ravetch and
Lanier, 2000). Further, ADCC can kill by direct lytic
action of enzymes such as granzyme and perforin.
Resistance to such enzymatic activity is another
potential avenue by which a tumor may evade killing
usually triggered by CD20 binding. Expression of
protease inhibitor 9 has now documented the existence
of this resistance mechanism (Bladergroen et al., 2002).
Thus, the nal decision of an effector cell to kill a target
or not can be affected by a complex interplay of signals,
including the nature of the interaction of the antibody
with FcgR that could be altered by either the antibody
or the FcgR, the presence of other activating or
inhibitory receptors and their ligands, or perturbation
of the downstream intracellular signals. With novel
signal inhibitors being studied, this latter possibility
becomes a testable method for indirectly overcoming a
number of potential antibody resistance mechanisms by
amplifying a weak antibody signal from the lymphoma
cell surface.
The preceding data indicate that FcgR are important
in antibody efcacy in mice and in patients, but do not
identify which cells are involved. NK cells appear to be
important, but would account only for the FcgRIIIA
data. The FcgRIIB data suggest a role for myeloid cells
and/or B cells. The myeloid growth factors G-CSF and
GM-CSF alter FcgR expression patterns and in some
systems enhance ADCC. Neutrophils could act by the
production of oxygen radicals, as well. The precise
contributions of the various effector cells to ADCC in
patients remain to be determined. This may be of critical
importance since the effector cell populations in patients
with lymphoma, especially after various treatment
modalities, may be variably depleted. Altered repertoires
of effector cells may lead to rituximab resistance, and
may account, for instance, for the improved response to
rituximab in patients with the small lymphocytic
subtype of lymphoma when rituximab is administered
as initial therapy (Hainsworth et al., 2002), in contrast
to the resistance observed in patients with this tumor
who are treated in the relapsed setting (McLaughlin
et al., 1998). The ability to restore the effector cell
compartments with cytokines such as interleukins (ILs)
2, 12 or 15 and myeloid growth factors, or potentially
with cellular replacement therapy, may also enhance
therapeutic antibody activity, as suggested by preclinical
data demonstrating that IL-2 can promote NK cell
development and enhance rituximab activity (Hooijberg
et al., 1995). Additional clinical trials are in progress
(Friedberg et al., 2002). Similarly, myeloid growth
factors in combination with rituximab are being
explored. Alternative approaches such as enhanced
effector cell activity induced by CpG oligonucleotides
are also being investigated (Warren et al., 2000).
The observation that some patients treated with
rituximab have either delayed responses or responses
that become more pronounced over a period of months
has led to several considerations. A cell cycle kinetic

Rituximab, mechanisms of action and resistance

MR Smith

7367

explanation is possible in that cells affected by antibody

binding may die only when they attempt to divide.
Persistent perturbations in cell membrane signaling by
alterations in lipid rafts are also theoretically possible
(discussed above). Altered immune function, however, is
also a possibility. Dying cells may release antigens
recognized by adaptive immune cells, leading to an
antilymphoma response. In vitro studies support such a
crosspriming effect (Selenko et al., 2002).
What can we learn from radioimmunotherapy results?
Radioactive anti-CD20 antibody therapy has been
approved for use in relapsed or refractory low-grade
or transformed lymphoma. In a randomized trial, this
agent, which consists of 90Yttrium bound to the murine
antibody parent of rituximab, administered along with
rituximab to help saturate blood and spleen CD20 sites,
had a higher rate of overall and complete responses
compared with rituximab alone (Witzig et al., 2002b).
Somewhat surprisingly, the time to disease progression
was not increased, suggesting intrinsic resistance of at
least a subpopulation of NHL cells. Also, since relapses
continue to occur following radioimmunotherapy, cure
is not generally attained and therefore resistance must
develop. In patients refractory to rituximab, overall
response rate to the radioimmunotherapy was similar to
that of patients who had not previously received
rituximab (Witzig et al., 2002a). Similar response rates
have been reported for a 131I-labeled murine CD20
antibody (Vose et al., 2000). The response to radioimmunotherapy in rituximab refractory patients suggests that the radioimmunoconjugate is acting by a
different mechanism, that is, targeted radiation therapy,
rather than the antibody mechanisms discussed above.
This suggests that resistance to radioimmunotherapy is
not due to aberrant CD20 signaling, or to deciencies of

ADCC or complement activation. Resistance to

radioimmunotherapy is not likely to be due to poor
access of the agent to lymphoma cells, since the path
length of the 90Y beta particle is about 10 mm. A
possible unifying basis for resistance to radiationinduced cell damage and to rituximab could be intrinsic
lymphoma cell resistance to apoptosis, for example,
altered apoptotic threshold. The paradox remains that
despite the initial response of most indolent lymphomas
to treatment, inevitably recurrence, implying a resistant
fraction of cells, occurs.

Summary
While therapy of lymphoma with rituxiamb is an
important addition to our therapeutic armamentarium,
not all patients respond and relapses are common. Until
we know the relative importance of the various potential
mechanisms of antibody-induced cell killing in patients,
it will be difcult to focus attention on the most
clinically important reasons for resistance. Ongoing
investigation into each of these areas will, therefore,
continue to be of importance, especially as it appears
likely that multiple mechanisms of action and of
resistance are operative, indicating that a multipronged
attack on resistance will be required. It will be important
to study patients serially, prior to and after rituximab
treatment and at the time of relapse, with regard to
properties of their lymphoma cells, looking for intrinsic
changes to account for resistance. In addition, parallel
studies of the innate and adaptive immune systems are
needed. The identication of resistance mechanisms
should lead to improved treatment, perhaps by antibody
engineering or cytokine or cellular therapies designed to
enhance effector actions.

References
Alas S, Ng C-P and Bonavida B. (2002). Clin. Cancer Res., 8,
836845.
Anderson KC, Bates MP, Slaughenhoupt BL, Pinkus GS,
Schlossman SF and Nadler LM. (1984). Blood, 63,
14241433.
Armitage JO and Weisenburger DD. (1998). J. Clin. Oncol.,
16, 27802795.
Bellosillo B, Villamor N, Lopez-Guillermo A, Marce S, Esteve
J, Campo E, Colomer D and Montserrat E. (2001). Blood,
98, 27712777.
Berinstein NL, Grillo-Lopez AJ, White CA, Bence-Bruckler I,
Maloney D, Czuczman M, Green D, Rosenberg J,
McLaughlin P and Shen D. (1998). Ann. Oncol., 9, 9951001.
Bladergroen B, Meijer CJLM, ten Berge RL, Hack CE,
Muris JJF, Dukers DF, Chott A, Kazama Y, Oudejans JJ,
van Berkum O and Kummer JA. (2002). Blood, 99,
232237.
Bubien JK, Zhou LJ, Bell PD, Frizzell RA and Tedder TF.
(1993). J. Cell Biol., 121, 11211132.
Byrd JC, Kitada S, Flinn IW, Aron JL, Pearson M, Lucas D
and Reed JC. (2002). Blood, 99, 10381043.
Byrd JC, Murphy T, Howard RS, Lucas MS, Goodrich A,
Park K, Pearson M, Waselenko JK, Ling G, Grever MR,

Grillo-Lopez AJ, Rosenberg J, Kunkel L and Flinn JW.

(2001). J. Clin. Oncol., 19, 21532164.
Campbell KS and Colonna M. (2001). Int. Rev. Immunol., 20,
333370.
Cartron G, Dacheux L, Salles G, Solal-Cellgny P and Bardos
P. (2002). Blood, 99, 754758.
Charles J and Foerster PAJ. (1999). Wintrobes Clinical
Hematology, Vol. 1, Chapter 44. Richard LG, John F, John
L, Frixos P, John G, George MR (eds). Williams & Wilkins:
Baltimore, MD, pp. 11911209.
Clynes RA, Towers TL, Presta LG and Ravetch JV. (2000).
Nat. Med., 6, 443446.
Colombat P, Salles G, Brousse N, Eftekhari P, Soubeyran P,
Delwail V, Deconinck E, Haioun C, Foussard C, Sebban C,
Stamatoullas A, Milpied N, Boue F, Taillan B, Lederlin P,
Najman A, Thieblemont C, Montestruc F, Mathieu-Boue A,
Benzohra A and Solal-Celigny P. (2001). Blood, 97, 101106.
Cragg MS, Morgan SM, Chan HTC, Morgan BP, Filatov AV,
Johnson PWM, French RR and Glennie MJ. (2003). Blood,
101, 10451052.
Czuczman MS, Grillo-Lopez AJ, McLaughlin P, White CA,
Saleh M, Gordon L, LoBuglio AE, Rosenberg J, Alkuzcweny B and Maloney D. (2001). Ann. Oncol., 12, 109114.
Oncogene

Rituximab, mechanisms of action and resistance

MR Smith

7368
Davis TA, Czerwinski DK and Levy R. (1999). Clin. Cancer
Res., 5, 611615.
Davis TA, Grillo-Lopez AJ, White CA, McLaughlin P,
Czuczman MS, Link BK, Maloney DG, Weaver RL,
Rosenberg J and Levy R. (2000). J. Clin. Oncol., 17,
31353143.
Deans JP, Kalt L, Ledbetter JA, Schieven GL, Bolen JB and
Johnson P. (1995). J. Biol. Chem., 270, 2263222638.
Deans JP, Li H and Polyak MJ. (2002). Immunology, 107,
176182.
Deans JP, Robbins SM, Polyak MJ and Savage JA. (1998).
J. Biol. Chem., 273, 344348.
Deans JP, Schieven GL, Shu GL, Valentine MA, Gilliland LA,
Aruffo A, Clark EA and Ledbetter JA. (1993). J. Immunol.,
15, 44944504.
Di Gaetano N, Xiao Y, Erba E, Bassan R, Rambaldi A,
Golay J and Introna M. (2001). Br. J. Haematol., 114,
800809.
Donin N, Jurianz K, Ziporen L, Schultz S, Kirschnk M and
Fishelson Z. (2003). Clin. Exp. Immunol., 131, 254263.
Fisher RI, Gaynor ER, Dahlberg S, Oken MM, Grogan TM,
Mize EM, Glick JH, Coltman Jr CA and Miller TP. (1993).
N. Engl. J. Med., 328, 10021006.
Foran JM, Gupta RK, Cunningham D, Popescu RA, Goldstone AH, Sweetenham JW, Pettengell R, Johnson PW,
Bessell E, Hancock B, Summers K, Hughes J, Rohatiner AZ
and Lister TA. (2000). Br. J. Haematol., 109, 8188.
Foran JM, Norton AJ, Micallef IN, Taussig DC, Amess JA,
Rohatiner AZ and Lister TA. (2001). Br. J. Haematol., 114,
881883.
Friedberg JW, Neuberg D, Gribben JG, Fisher DC, Canning
C, Koval M, Poor CM, Green LM, Daley J, Soiffer R, Ritz J
and Freedman AS. (2002). Br. J. Haematol., 117, 828834.
Ghetie M-A, Bright H and Vitetta ES. (2001). Blood, 97,
13921398.
Ghielmini M, Schmitz S-F, Cogliatti GS, Pichert G, Fey M,
Betticher D, Martinelli G, Peccatori F, Hess U, Stahel R,
Zucca E, Stupp R, Kovacsovics T, Helg C, Lohri A, Bargetzi
M, Vorobiof D, Cerny T, Group LW and Swiss Group for
Clinical Cancer Research (SAKK). (2002). Blood, 100, A604.
Golay JT, Clark EA and Beverley PC. (1985). J. Immunol.,
135, 37953801.
Golay J, Lazzari M, Facchinetti V, Bernasconi S, Borleri G
and Barbui T. (2001). Blood, 98, 33833389.
Golay J, Zaffaroni L and Vaccari T. (2000). Blood, 95,
39003908.
Greenlee RT, Murray T, Bolden S and Al E. (2000). CA
Cancer J. Clin., 50, 733.
Grillo-Lopez AJ. (2000). Semin. Oncol., 27, 916.
Hainsworth JD, Litchy S, Burris HA III, Scullin DC Jr, Corso
SW, Yardley DA, Morrissey L and Greco FA. (2002). J.
Clin. Oncol., 20, 42614267.
Harjunpaa A, Wiklund T, Collan J, Janes R, Rosenberg J, Lee
D, Grillo-Lopez A and Meri S. (2001). Leukemia Lymphoma, 42, 731738.
Harris NL, Jaffe ES, Stein H, Banks PM, Chan JK, Cleary
ML, Delsol G, DeWolf-Peeters C, Falini B and Gatter KC.
(1994). Blood, 84, 13611392.
Hofmeister JK, Cooney D and Coggeshall KM. (2000). Blood
Cells Mol. Dis., 26, 133143.
Hooijberg E, Sein JJ, van den Berk PCM, Hart AAM, van der
Valk MA, Kast WM, Melief CJM and Hekman A. (1995).
Cancer Res., 55, 26272634.
Keating MJ, Flinn IW, Jain V, Binet J-L, Hillman P, Byrd JC,
Albitar M, Brettman L, Santabarbara P, Wacker B and Rai
KR. (2002a). Blood, 99, 35543561.
Oncogene

Keating MJ, OBrien S and Albitar M. (2002b). Semin. Oncol.,

29, 7074.
Kennedy AD, Solga MD, Schuman TA, Chi AW, Lindorfer
MA, Sutherland WM, Foley PL and Taylor RP. (2003).
Blood, 101, 10711079.
Kennedy GA, Tey SK, Cobcroft R, Marlton P, Cull G,
Grimmett K, Thomson D and Gill D. (2002). Br. J.
Haematol., 119, 412416.
Koene HR, Kleijer M, Algra J, Roos D, von dem Borne AE
and de Haas M. (1997). Blood, 90, 11091114.
Leonard JP and Link BK. (2002). Semin. Oncol., 29, 8186.
Manches O, Lui G, Chaperot L, Gressin R, Molens J-P, Jacob
MC, Sotto J-J, Lerous D, Bens J-C and Plumas J. (2003).
Blood, 101, 949954.
Mathas S, Rickers A, Bommert K, Dorken B and Mapara
MY. (2000). Cancer Res., 60, 71707176.
McLaughlin P, Grillo-Lopez AJ, Link BK, Levy R, Czuczman
MS, Williams ME, Heyman MR, Bence-Bruckler I, White
CA, Cabanillas F, Jain V, Ho AD, Lister J, Wey K, Shen D
and Dallaire BK. (1998). J. Clin. Oncol., 8, 28252833.
OKeefe TL, Williams GT, Davies SL and Neuberger MS.
(1998). Immunogenetics, 48, 125132.
Osterborg A, Dyer MJ, Bunjes D, Pangalis GA, Bastion Y,
Catovsky D and Mellstedt H. (1997). J. Clin. Oncol., 15,
15671574.
Pedersen IM, Buhl AM, Klausen P, Geisler CH and Jurlander
J. (2002). Blood, 99, 13141319.
Petrie RJ and Deans JP. (2002). J. Immunol., 169, 28862891.
Pichert G, Schmitz SF, Hess U, Cerny T, Cogliatti SB,
Betticher D, Stupp R, Schmitter D, Stahel R and Ghielmini
M. (2001). Clin. Lymphoma, 1, 293297.
Piro LD, White CA, Grillo-Lopez AJ, Janakiraman N, Saven
A, Beck TM, Varns C, Shuey S, Czuczman M, Lynch JW,
Kolitz JE and Jain V. (1999). Ann. Oncol., 10, 655661.
Ravetch JV and Lanier LL. (2000). Science, 290, 8489.
Reff ME, Carner K, Chambers KS, Chinn PC, Leonard JE,
Raab R, Newman RA, Hanna N and Anderson DR. (1994).
Blood, 83, 435445.
Sacks D and Sher A. (2002). Nat. Immunol., 3, 10411047.
Selenko N, Majdic O, Jager U, Sillaber C, Stockl J and Knapp
W. (2002). J. Clin. Immunol., 22, 124130.
Shan D, Ledbetter JA and Press OW. (1998). Blood, 91,
16441652.
Shan D, Ledbetter JA and Press OW. (2000). Cancer Immunol.
Immunother., 48, 673683.
Tedder TF, Boyd AW, Freedman AS, Nadler LM and
Schlossman SF. (1985). J. Immunol., 135, 973979.
Treon SP, Mitsiades C, Mitsiades N, Young G, Doss D,
Schlossman R and Anderson KC. (2001). J. Immunother.,
24, 263271.
van der Kolk LE, de Haas M, Grillo-Lopez AJ, Baars JW and
van Oers M. (2002). Leukemia, 16, 693699.
Vose JM, Wahl RL, Saleh M and Al E. (2000). J. Clin. Oncol.,
18, 13161323.
Warren TL, Dahle CE and Weiner GJ. (2000). Clin.
Lymphoma, 1, 5761.
Weng W-K and Levy R. (2001). Blood, 98, 13521357.
Witzig TE, Flinn IW, Gordon L and Al E. (2002a). J. Clin.
Oncol., 20, 32623269.
Witzig TE, Flinn IW, Gordon LI, Emmanouilldes C, Czuczman MS, Saleh MN, Cripe L, Wiseman G, Olejnik T,
Multani PS and White CA. (2002b). J. Clin. Oncol., 20,
24532463.
Wu J, Edberg JC, Redecha PB, Bansal V, Guyre PM, Coleman
K, Salmon JE and Kimberly RP. (1997). J. Clin. Invest., 100,
10591070.